Artificial Neural Network Based Solar Radiation Estimation of Algeria Southwest Cities

Abstract

Solar radiation estimation is the most integral part of design and performance of solar energy applications. Our paper aims to develop an artificial neural networks-based model for predicting the daily global solar radiation in three cities (Bechar, Naâma and Tindouf) in the south-west region of Algeria. Models’ inputs are: average temperature, wind speed, relative humidity, atmospheric pressure, extraterrestrial solar irradiation, sunshine duration. Three ANN multilayer architectures connection are used with the Levenberg-Marquardt algorithm for training. Efficiency of models was assessed using statistical tests including, correlation coefficient (R), root mean squared error (RMSE), mean bias error (MBE) and mean absolute percentage error (MAPE).The results during five years showed that, the Cascade-forward Neural Network (CFNN) and Feed-forward neural network (FFNN) models gives much better forecast of daily global solar radiation in the three Saharan cities. The models developed can be used for design and sizing solar energy systems, where radiation measuring stations are scarce in Algeria.